A Novel Framework to Process the Quantity and Quality of User Behavior Data in Recommender Systems

2016 ◽  
pp. 231-243 ◽  
Author(s):  
Penghua Yu ◽  
Lanfen Lin ◽  
Yuangang Yao
Keyword(s):  
Recommender Systems ◽  
User Behavior

Web Content Recommendation Methods Based on Reinforcement Learning

2011 ◽  
pp. 2353-2380
Author(s):  
Nima Taghipour ◽  
Ahmad Kardan
Keyword(s):  
Reinforcement Learning ◽  
Recommender Systems ◽  
Information Overload ◽  
User Behavior ◽  
Web Content ◽  
Web Usage ◽  
Content Recommendation ◽  
Conceptual Relationships ◽  
The Web

Information overload is no longer news; the explosive growth of the Internet has made this issue increasingly serious for Web users. Recommender systems aim at directing users through this information space, toward the resources that best meet their needs and interests. In this chapter we introduce our novel machine learning perspective toward the web recommendation problem, based on reinforcement learning. Our recommendation method makes use of the web usage and content data to learn a predictive model of users’ behavior on the web and exploits the learned model to make web page recommendations. Unlike other recommender systems, our system does not use the static patterns discovered from web usage data, instead it learns to make recommendations as the actions it performs in each situation. In the proposed method we combined the conceptual and usage information in order to gain a more general model of user behavior and improve the quality of web recommendations. A hybrid web recommendation method is proposed by making use of the conceptual relationships among web resources to derive a novel model of the problem, enriched with semantic knowledge about the usage behavior. The method is evaluated under different settings and it is shown how this method can improve the overall quality of recommendations.

PageRank Algorithm-Based Recommender System Using Uniformly Average Rating Matrix

2019 ◽  
pp. 99-112
Author(s):  
Bathrinath S. ◽  
Saranyadevi S. ◽  
Thirumalai Kumaran S. ◽  
Saravanasankar S.
Keyword(s):  
Recommender Systems ◽  
Large Scale ◽  
User Behavior ◽  
Personalized Recommendation ◽  
Pagerank Algorithm ◽  
Large Scale Networks ◽  
Matrix Construction ◽  
Item Item ◽  
New Item

Applications of web data mining is the prediction of user behavior with respect to items. Recommender systems are being applied in knowledge discovery techniques to the problem of making decisions on personalized recommendation of information. Traditional CF approaches involve the amount of effort increases with number of users. Hence, new recommender systems need to be developed to process high quality recommendations for large-scale networks. In this chapter, a model for UAR matrix construction method for item rank calculations, a Page Rank-based item ranking approach are proposed. The analysis of various techniques for computing item-item similarities to identify relationship between the selected items and to produce a qualified recommendation for users to acquire the items as their wish. As a result, the new item rank-based approaches improve the quality of recommendation outcome. Results show that the proposed UAR method outperforms than the existing method. The same method is applied for the large real-time rating dataset like Movie Lens.

Web Content Recommendation Methods Based on Reinforcement Learning

2010 ◽  
pp. 222-249
Author(s):  
Nima Taghipour ◽  
Ahmad Kardan
Keyword(s):  
Reinforcement Learning ◽  
Recommender Systems ◽  
Information Overload ◽  
User Behavior ◽  
Web Content ◽  
Web Usage ◽  
Content Recommendation ◽  
Conceptual Relationships ◽  
The Web

Information overload is no longer news; the explosive growth of the Internet has made this issue increasingly serious for Web users. Recommender systems aim at directing users through this information space, toward the resources that best meet their needs and interests. In this chapter the authors introduce their novel machine learning perspective toward the Web recommendation problem, based on reinforcement learning. Our recommendation method makes use of the Web usage and content data to learn a predictive model of users’ behavior on the Web and exploits the learned model to make Web page recommendations. Unlike other recommender systems, this system does not use the static patterns discovered from Web usage data, instead it learns to make recommendations as the actions it performs in each situation. In the proposed method the authors combined the conceptual and usage information in order to gain a more general model of user behavior and improve the quality of web recommendations. A hybrid Web recommendation method is proposed by making use of the conceptual relationships among Web resources to derive a novel model of the problem, enriched with semantic knowledge about the usage behavior. The method is evaluated under different settings and it is shown how this method can improve the overall quality of recommendations.

scholarly journals Critique on Natural Noise in Recommender Systems

2021 ◽  
Vol 15 (5) ◽  
pp. 1-30
Author(s):  
Wissam Al Jurdi ◽  
Jacques Bou Abdo ◽  
Jacques Demerjian ◽  
Abdallah Makhoul
Keyword(s):  
Recommender Systems ◽  
User Behavior ◽  
Cold Start ◽  
User Experiences ◽  
Data Sparsity ◽  
Natural Noise ◽  
E Learning ◽  
Management Techniques ◽  
Noise Management

Recommender systems have been upgraded, tested, and applied in many, often incomparable ways. In attempts to diligently understand user behavior in certain environments, those systems have been frequently utilized in domains like e-commerce, e-learning, and tourism. Their increasing need and popularity have allowed the existence of numerous research paths on major issues like data sparsity, cold start, malicious noise, and natural noise, which immensely limit their performance. It is typical that the quality of the data that fuel those systems should be extremely reliable. Inconsistent user information in datasets can alter the performance of recommenders, albeit running advanced personalizing algorithms. The consequences of this can be costly as such systems are employed in abundant online businesses. Successfully managing these inconsistencies results in more personalized user experiences. In this article, the previous works conducted on natural noise management in recommender datasets are thoroughly analyzed. We adequately explore the ways in which the proposed methods measure improved performances and touch on the different natural noise management techniques and the attributes of the solutions. Additionally, we test the evaluation methods employed to assess the approaches and discuss several key gaps and other improvements the field should realize in the future. Our work considers the likelihood of a modern research branch on natural noise management and recommender assessment.

scholarly journals User Identification in the Process of Web Usage Data Preprocessing

2019 ◽  
Vol 14 (09) ◽  
pp. 21 ◽  
Author(s):  
Jozef Kapusta ◽  
Michal Munk ◽  
Dominik Halvoník ◽  
Martin Drlík
Keyword(s):  
User Behavior ◽  
Web Server ◽  
Web Browser ◽  
Ip Address ◽  
User Session ◽  
Log Files ◽  
The Us ◽  
Log File ◽  
Default Form

If we are talking about user behavior analytics, we have to understand what the main source of valuable information is. One of these sources is definitely a web server. There are multiple places where we can extract the necessary data. The most common ways are to search for these data in access log, error log, custom log files of web server, proxy server log file, web browser log, browser cookies etc. A web server log is in its default form known as a Common Log File (W3C, 1995) and keeps information about IP address; date and time of visit; ac-cessed and referenced resource. There are standardized methodologies which contain several steps leading to extract new knowledge from provided data. Usu-ally, the first step is in each one of them to identify users, users’ sessions, page views, and clickstreams. This process is called pre-processing. Main goal of this stage is to receive unprocessed web server log file as input and after processing outputs meaningful representations which can be used in next phase. In this pa-per, we describe in detail user session identification which can be considered as most important part of data pre-processing. Our paper aims to compare the us-er/session identification using the STT with the identification of user/session us-ing cookies. This comparison was performed concerning the quality of the se-quential rules generated, i.e., a comparison was made regarding generation useful, trivial and inexplicable rules.

scholarly journals A New Similarity Measure for User-based Collaborative Filtering in Recommender Systems

2015 ◽  
Vol 14 (9) ◽  
pp. 6118-6128 ◽  
Author(s):  
T. Srikanth ◽  
M. Shashi
Keyword(s):  
Collaborative Filtering ◽  
Recommender Systems ◽  
Similarity Measure ◽  
Recommender System ◽  
Distance Measure ◽  
Current User ◽  
Challenging Problem ◽  
Popular Approach ◽  
Similarity Calculation

Collaborative filtering is a popular approach in recommender Systems that helps users in identifying the items they may like in a wagon of items. Finding similarity among users with the available item ratings so as to predict rating(s) for unseen item(s) based on the preferences of likeminded users for the current user is a challenging problem. Traditional measures like Cosine similarity and Pearson correlation’s correlation exhibit some drawbacks in similarity calculation. This paper presents a new similarity measure which improves the performance of Recommender System. Experimental results on MovieLens dataset show that our proposed distance measure improves the quality of prediction. We present clustering results as an extension to validate the effectiveness of our proposed method.

scholarly journals Syn Rec: A Prediction Technique using Collaborative Filtering and Synergy Score

2019 ◽  
Vol 8 (5S3) ◽  
pp. 457-463
Keyword(s):  
Social Interactions ◽  
Collaborative Filtering ◽  
Recommender Systems ◽  
Recommender System ◽  
Effective Technique ◽  
Novel Technique ◽  
Trust Relation ◽  
Prediction Technique ◽  
User Trust

Recommender systems are techniques designed to produce personalized recommendations. Data sparsity, scalability cold start and quality of prediction are some of the problems faced by a recommender system. Traditional recommender systems consider that all the users are independent and identical, its an assumption which leads to a total ignorance of social interactions and trust among user. Trust relation among users ease the work of recommender systems to produce better quality of recommendations. In this paper, an effective technique is proposed using trust factor extracted with help of ratings given so that quality can be improved and better predictions can be done. A novel-technique has been proposed for recommender system using film-trust dataset and its effectiveness has been justified with the help of experiments.

Trust-Aware Recommender Systems for Open and Mobile Virtual Communities

2010 ◽  
pp. 326-345
Author(s):  
N. Sahli ◽  
G. Lenzini
Keyword(s):  
Recommender Systems ◽  
Virtual Communities ◽  
Multi Agent Systems ◽  
Artificial Intelligent ◽  
Agent Based ◽  
Agent Systems ◽  
Multi Agent

This chapter surveys and discusses relevant works in the intersection among trust, recommendations systems, virtual communities, and agent-based systems. The target of the chapter is showing how, thanks to the use of trust-based solutions and artificial intelligent solutions like that understanding agents-based systems, the traditional recommender systems can improve the quality of their predictions. Moreover, when implemented as open multi-agent systems, trust-based recommender systems can efficiently support users of mobile virtual communities in searching for places, information, and items of interest.

DSS for Web Mining Using Recommendation System

2017 ◽  
pp. 22-34 ◽  
Author(s):  
Varaprasad Rao M ◽  
Vishnu Murthy G
Keyword(s):  
Recommender Systems ◽  
Web Mining ◽  
Recommendation System ◽  
User Behavior ◽  
Web Content ◽  
Web Structure ◽  
Web Structure Mining ◽  
Content Mining ◽  
Computer Based ◽  
The Many

Decision Supports Systems (DSS) are computer-based information systems designed to help managers to select one of the many alternative solutions to a problem. A DSS is an interactive computer based information system with an organized collection of models, people, procedures, software, databases, telecommunication, and devices, which helps decision makers to solve unstructured or semi-structured business problems. Web mining is the application of data mining techniques to discover patterns from the World Wide Web. Web mining can be divided into three different types – Web usage mining, Web content mining and Web structure mining. Recommender systems (RS) aim to capture the user behavior by suggesting/recommending users with relevant items or services that they find interesting in. Recommender systems have gained prominence in the field of information technology, e-commerce, etc., by inferring personalized recommendations by effectively pruning from a universal set of choices that directed users to identify content of interest.

Neighborhood Evaluation in Recommender Systems Using the Realization Based Entropy Approach

2014 ◽  
Vol 1 (4) ◽  
pp. 34-50
Author(s):  
Roee Anuar ◽  
Yossi Bukchin ◽  
Oded Maimon ◽  
Lior Rokach
Keyword(s):  
Recommender Systems ◽  
Entropy Measure ◽  
Continuous Data ◽  
System Evaluation ◽  
Rating Data ◽  
New Approach ◽  
Selection Of ◽  
Classical Entropy ◽  
Better Than

The task of a recommender system evaluation has often been addressed in the literature, however there exists no consensus regarding the best metrics to assess its performance. This research deals with collaborative filtering recommendation systems, and proposes a new approach for evaluating the quality of neighbor selection. It theorizes that good recommendations emerge from good selection of neighbors. Hence, measuring the quality of the neighborhood may be used to predict the recommendation success. Since user neighborhoods in recommender systems are often sparse and differ in their rating range, this paper designs a novel measure to asses a neighborhood quality. First it builds the realization based entropy (RBE), which presents the classical entropy measure from a different angle. Next it modifies the RBE and propose the realization based distance entropy (RBDE), which considers also continuous data. Using the RBDE, it finally develops the consent entropy, which takes into account the absence of rating data. The paper compares the proposed approach with common approaches from the literature, using several recommendation evaluation metrics. It presents offline experiments using the Netflix database. The experimental results confirm that consent entropy performs better than commonly used metrics, particularly with high sparsity neighborhoods. This research is supported by The Israel Science Foundation, Grant #1362/10. This research is supported by NHECD EC, Grant #218639.

Sign in / Sign up

Export Citation Format

Share Document